Malaria Vaccine: Candidate Antigens, Mechanisms, Constraints and Prospects

Scand. J. Immunol. 56, 327±343, 2002 REVIEW Malaria Vaccine: Candidate Antigens, Mechanisms, Constraints and Prospects L. J. M. CARVALHO,*C. T. DANIEL-RIBEIRO*& H. GOTOy *Laboratory of Malaria Research, Department of Immunology, WHO Collaborating Centre for Research and Training in the Immunology of Parasitic Diseases, Instituto Oswaldo Cruz, Rio de Janeiro, RJ; and ylaboratory of Soroepidemiology and Immunobiology, SaÄo Paulo Institute of Tropical Medicine and Department of Preventive Medicine, University of SaÄo Paulo, SaÄo Paulo, SP, Brazil (Received 15 March 2002; Accepted in revised form 1 July 2002) Carvalho LJM, Daniel-Ribeiro CT, Goto H. Malaria Vaccine: Candidate Antigens, Mechanisms, Constraints and Prospects. Scand J Immunol 2002;56:327±343 More than 30 years after the first report of successful vaccination against malaria using radiationattenuated sporozoites, an effective malaria vaccine is not yet available. However, field and experimental data indicate that it can be developed. An astonishing amount of data has accumulated concerning parasite biology, host±parasite interactions, immunity and escape mechanisms, targets and modulators of immune responses. Nevertheless, so far this knowledge has not been enough to make us understand how to properly manipulate the whole system to build an effective vaccine. In this article, we describe candidate antigens, mechanisms, targets and trials performed with potential malaria vaccines and discuss the approaches, in vivo and in vitro models, constraints and how technologies such as DNA vaccination, genomics/proteomics and reverse immunogenetics are providing exciting results and opening new doors to make malaria vaccine a reality. Dr L. J. M. Carvalho, Laboratory of Malaria Research, Department of Immunology, Instituto Oswaldo Cruz, Fiocruz, Av. Brazil 4365, Manguinhos, 21045-900 Rio de Janeiro, RJ, Brazil. E-mail: leojmc@ioc.fiocruz.br THE MALARIA PROBLEM Mankind has stepped into the 21st century and diseases such as malaria still represent a major threat to populations in many parts of the world. The exact extent of the malaria problem is not known, but several estimates provide a gloomy picture of the situation [1]. It is estimated that between 400 and 900 million febrile episodes occur every year only in African children, with a minimum of 750,000 deaths (probably up to 3 million). In addition, nonsevere cases yet cause considerable morbidity in acute or chronic disease, with serious socio-economical consequences. Sub- Saharan Africa is the most affected region in the world, but malaria is also a serious problem in several other places, such as South-East Asia, Oceania, Middle East and Latin America. Historically, in the late 1940s, there was great optimism in the fight against malaria, mainly owing to the introduction of dichlorodiphenyltrichloroethane (DDT) for vector control and of chloroquine as a very efficient antimalarial drug. These and other available control tools prompted the World Health Organization (WHO) to launch a campaign for complete malaria eradication. The campaign was very successful in places such as Mediterranean countries and even many regions in the tropics. In Brazil, for instance, the number of cases per year dropped from nearly 6 million in the 1940s to around 37,000 in 1962 [2] and became restricted to the Amazon region. But since then, malaria has seen resurgence and/ or is spreading in many areas; in the Amazon region of Brazil the number of cases per year increased from nearly 37,000 in 1962 to around 600,000 in the late 1990s. Indeed, already in the 1960s it became clear that eradication was not feasible, and the WHO strategy was switched, aiming to control rather than to eradicate. Environmental conditions, population habits and living conditions, migratory movements of people to endemic areas, regional development projects, resistance of parasites to drugs and of mosquitoes to DDT, among other factors, greatly favoured the maintenance of malaria in the endemic regions. All the difficulties concerning malaria control justify the search and adoption of new tools and measures to minimize the impact of malaria on the # 2002 Blackwell Science Ltd

328 L. J. M. Carvalho et al. affected populations. One strategy is the development of a malaria vaccine. BIOLOGY Four species of Plasmodium cause human malaria, the most prevalent being Plasmodium falciparum and Plasmodium vivax and less frequently Plasmodium malariae and Plasmodium ovale [3]. Briefly, the plasmodial life cycle in man starts with the inoculation of a few sporozoites through the bite of an Anopheles mosquito. Within minutes, sporozoites invade the hepatocyte and start their development, multiplying by schizogony (liver stage). In the case of P. vivax and P. ovale, some sporozoites may differentiate into hypnozoites, responsible for late relapses of the infection. After a period of 5±14 days ± depending on the plasmodial species ± schizonts develop into thousands of merozoites that are freed into the bloodstream and invade the red blood cells (RBCs), initiating the blood stage. In the RBC, each merozoite develops into a trophozoite that matures and divides, generating a schizont that, after fully matured, gives rise to up to 32 merozoites within 42±72 h, depending on the plasmodial species. The merozoites, freed into the bloodstream, will invade other RBC, maintaining the cycle. Some merozoites, after invading a RBC, develop into sexual forms ± the male or female gametocytes. If a female Anopheles mosquito takes its blood meal and ingests the gametocytes, it will become infected. In the mosquito gut, the male gametocyte fertilizes the female gametocyte, generating the ookinete, which binds to and passes through the gut wall, remains attached to its external face and transforms into the oocyst. The oocyst will divide by sporogony, giving rise to thousands of sporozoites that will be freed in the body cavity of the mosquito and eventually migrate to its salivary gland, where they will maturate, becoming capable of starting a new infection in humans when the mosquito bites the host for a blood meal. MALARIA VACCINE: REASONS FOR HOPE The hope for developing a malaria vaccine is based on field and experimental observations, showing that immune protection against malaria can be achieved: (a) individuals living in areas of intense transmission naturally acquire clinical immunity, first against severe disease and then against clinical manifestations; partial antiparasite immunity is also developed, and adults may have parasites in the blood but usually at very low densities without causing symptoms [4, 5]; (b) immunity can be passively transferred from immune to nonimmune individuals through the administration of immunoglobulins, showing that protection against blood-stage infection is largely mediated by antibodies [6, 7]; (c) immunization of humans, primates and mice with radiation-attenuated sporozoites or with recombinant antigens can induce partial or sterile (100% effective) antiparasite immunity [8]. Nevertheless, the belief that a malaria vaccine can be developed has been often questioned by the fact that natural immunity takes many years to be acquired, and this happens only under continuous and heavy contact with the parasite, as in holo- and hyperendemic areas; in addition, immunity is partial (nonsterile) and short lived, if continuous boosting is not present. For most of the pathogens against which vaccines have been developed, such as measles and mumps, nonimmune individuals surviving an infection develop strong long-lived sterile immunity. In those cases, the vaccine just does what the nature is able to do. In the case of malaria, a vaccine has to be more efficient than nature. Could that be possible? Some observations indicate that it could be. First, the fact that only adults are immune to malaria in holo- and hyperendemic areas seems to be related to the host age or, in a better context, to its developmental state, rather than a consequence of long-term exposure [9, 10]. Although children are not able to develop efficient protective immunity against the parasite, they can quickly develop immunity against severe disease [11]. In addition, the increased resistance to malaria has been shown to be directly related to pubertal hormone levels [12]. These observations indicate that immunity is achieved in a more efficient manner than previously thought. It is largely dependent on how the response is modulated, and it may be easier to develop an efficient vaccine for adults than for children. However, as children must be the main focus for vaccination in many communities, it is of key importance that the mechanisms behind their low ability to develop immunity be clarified, so that a vaccine may be effective irrespective of the vaccinee's age. Second, strong protective immunity is induced after a single inoculation of radiation-attenuated sporozoites but not after repeated natural challenge with normal sporozoites. One of the possible reasons for this phenomenon is that irradiated sporozoites invade hepatocytes and express liverstage antigens, and, because they do not progress beyond the liver stage, they become a source of persistent intrahepatic antigens [13]. It is reasonable to think that a synthetic vaccine having the same property might also induce protective immunity. These observations raise optimism in the expectation that an artificial intervention ± a vaccine ± may be more efficient than nature in inducing protective immunity against malaria. CONSTRAINTS Many of the currently available effective vaccines were developed on an empirical or semiempirical basis, even before the establishment of some landmark concepts of immunology, such as the clonal selection theory of Burnet and the role of the thymus and T lymphocytes [14, 15]. For malaria, semiempirical approaches may work as well, as it has been with attenuated sporozoites. However, more rational and potentially effective approaches will need a substantial improvement,

Malaria Vaccines 329 along with our knowledge of mechanisms behind the acquisition of protective immunity, their interactions, how to induce and modulate them, and how to reliably study them. Our failure so far may be related to the fact that immunity is complex and multifactorial, and linear research may not be able to solve the puzzle [16]. Some of the difficulties to develop a malaria vaccine involve essentially biological problems imposed by a parasite having a complex life cycle and establishing complex interactions with its host. Each stage of the parasite development is characterized by different sets of expressed antigens, eliciting different types of immune responses (Fig. 1) [17]. Despite ± or maybe because of ± displaying a myriad of antigenic molecules, the immune response against them is often ineffective. One of the reasons is the extensive polymorphism of plasmodial antigens, leading to differences in their antigenic properties in different isolates [18]. In addition, for some proteins, there are sets of different coding genes producing variant proteins that would be sequentially expressed during infection in response to immune pressure [19]. Host genetic restriction of antigen recognition, mediated by major histocompatibility complex (MHC) haplotypes, may also negatively influence the response against plasmodial antigens [20±22], although genetic immunization seems to circumvent this problem [23]. The parasite may also have deleterious influence on the host immune system ± for instance, by causing polyclonal B-cell activation [24] and immunodepression [25]. Infected RBCs may also inhibit the maturation of dendritic cells, reducing their capacity to stimulate T cells [26]. Some technical limitations are also involved in preventing malaria vaccine development. The lack of reliable correlates of protection and of in vitro and/or in vivo surrogate assays is a major one (see below). Another limitation is the poor availability of adjuvants able to significantly strengthen the response to a given antigen without causing deleterious reactions or side effects [27]. In what concerns political and financial aspects, malaria is a disease of poor people, and malaria vaccine research raises little interest in private groups such as pharmaceutical companies, it being largely supported by public funding. Recent advances are encouraging, with initiatives of public±private partnerships and antimalarial programmes of private companies [28], besides major funding from nonpublic sources [29, 30]. Among all the drawbacks in malaria vaccine development, we would like to emphasize three aspects. First, the acquisition of immunity in malaria is still far from being a wellunderstood phenomenon. Several mechanisms of protection have been described for liver and blood stages, and many antigens inducing such mechanisms have been identified. However, so far no reliable correlates of protection have been identified, often turning malaria vaccine research into an essentially empirical or semiempirical approach. A given mechanism described as important ± for instance, cytotoxic T lymphocyte (CTL) activity in liver-stage immunity ± is sometimes detected in nonprotected individuals and not detected in protected ones [31]. Similarly, candidate antigens Fig. 1. Life cycle of Plasmodium depicting the potential targets and mechanisms of protective immune responses. Major candidate antigens for malaria vaccine derive from the six listed targets are presented (modified from the figure published by Hoffman and Miller in Ref. [17]).

330 L. J. M. Carvalho et al. are often identified on the basis of their reactivity with sera or stimulation of T cells from immune individuals, after the screening of expression clones from Plasmodium DNA libraries. The fact that a given antigen is recognized by a large proportion of sera or T cells of immune individuals but being poorly or not recognized by exposed nonimmune ones in epidemiological surveys may indicate an association between the response against that antigen and protection. This rationale, despite standing on the basis of the identification of most of the current candidate antigens, is of limited value, because the presence of antibodies or T cells specific for agivenantigenmaybesimplyamarkeraccompanyingthe acquisition of immunity rather than be the elements actively involved in protection. This shows that immunity to malaria is not just a question of the mechanisms acting but rather of how the many complexes and subtle interactions are working for modulating the response to make it effective. Second, in vitro assays are often explored to support an antigen as a candidate. However, these assays are of limited predictive value. As well-defined correlates of protection have not been identified, there is no available in vitro assay that may function as a surrogate marker of protection in malaria, whether considering immunity against blood or preerythrocytic stages. For instance, the demonstration that antibodies against a given antigen have an inhibitory effect on parasite growth in vitro is often considered as a clue or a proof that this antigen elicits a protective response. However, some antibodies highly inhibitory in vitro may have no effect in vivo [32], and vice versa [7]. In the case of liver stage, cellular immunity is regarded as essential, acting mainly through cytokine (interferon-g (IFN-g)) release and its downstream effects such as nitric oxide (NO) production [33]. In humans, studies of cellular immunity are restricted to the use of peripheral blood cells, and this brings about important restrictions, because these cells may not be representative of the tissue pool [34]. For instance, assuming that a hypothetical liver-stage antigen induces IFN-g-producing CD8 T cells with a strong liver-homing phenotype, which would have very efficient parasite-killing properties and low capacity of recirculation in the peripheral blood, it is likely that this strong malaria vaccine candidate would not even be identified. On the other hand, cells with weak liver-homing phenotype and consequently with less-efficient parasite killing ability might be selected. This also raises the question of the relevance of certain analytical procedures. Studying CTL activity, IFN-g production and other functional analysis of cellular activity using peripheral blood cells may not bring a reliable picture of those activities in the sites where they are important, especially the liver. The lack of reliable surrogate markers of protection poses a major problem for the screening and evaluation of vaccine candidates. A third restriction is how suitable are the available experimental animals to model human malaria. The murine models in particular are very commonly used to study host±parasite interactions and mechanisms of immunity. However, murine malaria is quite different from human malaria, and the mechanisms of immunity acting in such models may have no relevance for humans. Nevertheless, some researchers argue that the murine model often closely predicts the outcome in humans [35]. The New World monkeys Aotus and Saimiri are the primate models recommended by WHO for preclinical trials of malaria vaccines [36], mainly because they can develop reproducible infections by both P. falciparum and P. vivax. However, while their usefulness in evaluating immunogenicity and efficacy of potential candidates is well established and widely accepted, their assignment as a critical path in the vaccine development has been a matter of discussion [37, 38]. In fact, Gray Heppner and colleagues strongly criticize these models, stating that owing to considerable differences between humans and these primates, screening vaccine candidates using these models might lead to situations where poor candidates might be selected for human trials, whilst potentially effective antigens for humans might be discarded. But similar argumentation might be imposed to all the other steps in the development and preclinical testing of malaria vaccine candidates (see the section below). Under very strict considerations, none of those steps would be on the critical path. In conclusion, we would like to stress that most of the data supporting the current vaccine candidate antigens rely on the three above approaches, i.e. epidemiological associations between protection and antigen recognition by exposed individuals, antiparasitic effect in vitro and experiments in animal models. As their relevance is far from being well established and consensual, they mostly function as a guide rather than as a secure way for finding and studying potential candidate antigens. It means that many times the bet on a new antigen is a shot in the dark. MALARIA VACCINE DEVELOPMENT: FROM BENCH TO BUSH Despite all constraints, malaria vaccine research is very active. In this section, we will briefly describe the steps toward the development of a conventional subunit malaria vaccine. Different approaches such as reverse immunogenetics and reverse vaccinology will be dealt with in separate sections. Conventional subunit vaccine development may be roughly divided in three phases: research and development (R&D), preclinical testing and clinical trials. The first two concern laboratory and experimental work, with the identification, cloning and characterization of antigens targeted by potentially protective immune mechanisms (verified in vitro and in vivo in murine models) and then their evaluation (safety, immunogenicity and efficacy) in nonhuman primate models, applying Good Laboratory Practice (GLP) procedures to guarantee the quality of the products potentially deliverable for future trials [39]. GLP products may be evaluated

Malaria Vaccines 331 for safety, immunogenicity and efficacy in nonhuman primates (Aotus, Saimiri and, eventually, the chimpanzee), using different adjuvants, delivery systems, protocols and dosages. Once the best formulations and protocols have been defined, they can be forwarded to human clinical trials. For human trials, procedures must comply with the higher exigencies of Good Manufacturing Practice (GMP) standards, with the delivery of clinical-grade products. Clinical trials are divided in four phases. Phase I is intended to evaluate safety, immunogenicity and the types of immune response elicited in nonimmune volunteers living outside endemic areas or in limited number of exposed individuals. Efficacy is evaluated in Phase II: Phase IIa involves nonimmune nonexposed volunteers, and challenge is performed through the bites of laboratory-reared sporozoite-carrying mosquitoes, and efficacy is evaluated as the ability to avoid or delay infection. Owing to ethical requirements, the trial end point is reached with the detection of a single bloodstage form, when treatment is promptly established. This is a considerable limitation especially for evaluating blood-stage vaccines, which may be very effective in eliminating the parasite but only after allowing a limited growth. Some of us have observed this pattern in preclinical trials of two blood-stage antigens in Saimiri monkeys (unpublished data). In this work, high prechallenge antibody titres were able to keep parasitaemia at low levels, allowing the infection to strongly boost the immune response, which then eliminated the parasite. So, early treatment in a Phase IIa trial might overlook this kind of effect. Phase IIb targets a limited number of volunteers from endemic areas, who are exposed to natural challenge. Protection may be assessed, for instance, through the ability of vaccinated individuals to get no or less infections or to have longer intervals between infections, as compared with a control group. Candidate formulations going successfully through these stages may be further evaluated in Phase III trial, where the vaccine is delivered to populations in a number of endemic regions displaying different epidemiological characteristics, and the end point may be a statistically significant decrease in the incidence of malaria or of severe disease in relation to a nonvaccinated population under the same epidemiological setting. A successful outcome will lead the vaccine to Phase IV, where it is administered in much larger populations, most likely together with other already registered vaccines, such as within the context of the WHO-oriented Extended Programme of Immunization (EPI), where its long-term efficacy and possible interferences with other vaccines may be properly evaluated. SPf66, developed in Colombia by Manuel Patarroyo, has been the only malaria vaccine candidate to go up to Phase III trials (see `Multistage approaches'). A number of other candidate vaccines have been evaluated in Phase I or II trials, and several antigens have been identified and are being evaluated in preclinical trials. We will describe how researchers have been approaching malaria vaccine development, exploring potential targets and protective mechanisms acting in the different stages of the parasite cycle, both in the human or in the mosquito host. The focus will be on P. falciparum antigens. Information on P.vivaxvaccines can be found in a recent review [40]. SIX TARGETS Considering the whole parasite life cycle, there are essentially six targets for a malaria vaccine (Fig. 1): (1) sporozoites; (2) liver stages; (3) merozoites; (4) infected RBC; (5) parasite toxins; (6) sexual stages. Table 1 summarizes the main candidate antigens of each stage identified so far, and Table 2 provides an overview of relevant preclinical and clinical trials conducted with some of these antigens. In this section, we will stick to more conventional approaches. DNA vaccines, reverse immunogenetics and genomics/microarrays will be dealt with in separate sections. Sporozoite The sporozoite remains in the bloodstream for a very short period of time before invading a hepatocyte, making it an unlikely target for an effective protective response. However, it was using radiation-attenuated sporozoites that Nussenzweig Table1. Main vaccine candidates from the different phases of Plasmodium life cycle Targets Candidate antigens Sporozoite Circumsporozoite protein (CSP) Thrombospondin-related adhesive protein (TRAP) Sporozoite and liver-stage antigen (SALSA) Sporozoite threonine- and asparagine-rich protein (STARP) Liver stage CSP Liver-stage antigen (LSA)-1 and -3 SALSA STARP Merozoite Merozoite surface protein (MSP)-1, -2, -3, -4 and -5 Erythrocyte-binding antigen (EBA)-175 Apical membrane antigen (AMA)-1 Rhoptry-associated protein (RAP)-1 and -2 Acidic-basic repeat antigen (ABRA) Duffy-binding protein (DBP) (Plasmodium vivax) Blood stage Ring erythrocyte surface antigen (RESA) Serine-rich protein (SERP) Erythrocyte membrane protein (EMP)-1, -2 and -3 Glutamate-rich protein (GLURP) Toxins Glycosilphosphatidylinositol (GPI) Sexual stages Ps25, Ps28, Ps48/45 and Ps230

332 L. J. M. Carvalho et al. Table2. Selected trials with malaria vaccine candidate antigens at preclinical (nonhuman primates) or clinical phases Candidate Phase Formulation and scheme Outcome References Single antigen constructs NANP Phase III Recombinant protein No significant protection [47] RTS,S/AS02 Phase IIa Fusion protein of recombinant CSP and HBsAg Initial sterile protection in six of seven [48] volunteers, but short lived RTS,S/AS02 Phase IIb Fusion protein of recombinant CSP and HBsAg 34% efficacy, but short-lived protection [49] LSA-3 Preclinical (chimpanzee) Lipopeptides in saline or polypeptides in ISA51 or SBAS2 adjuvants Full protection in six and partial protection in two of nine immunized animals MSP-1 Phase I Recombinant MSP-1 19 with tetanus toxoid Adverse local reactions [66] MSP-1 Preclinical (Aotus) Recombinant MSP-1 19 with seven Protection achieved only with Freund's adjuvant [64] different adjuvants Full-length recombinant protein in CFA MSP-3 Preclinical (Aotus and Saimiri) followed by ISA51 adjuvants; MSP3 212 380 recombinant protein in three different adjuvants EBA-175 Preclinical (Aotus) Recombinant EBA-175 or its DNA followed by challenge with recombinant protein Protection of six of seven Aotus; protection with MSP3 212 380 in AS02 adjuvant correlated with antibody titres Protection (some monkeys did not require treatment and others had lower peaks of parasitaemia) [52] [65] and authors' unpublished data AMA-1 Preclinical (Saimiri) Plasmodium fragile AMA-1 Partial protection against Plasmodium falciparum [78] RAP-1 and RAP-2 Preclinical (Saimiri) Native and recombinant proteins in Partial protection [79] Freund's and Montanide adjuvants GLURP Preclinical (Saimiri) Recombinant protein GLURP 27 500 One of three monkeys self-cleared infection Authors' unpublished data in alum adjuvant Multiantigen constructs SPf66 Phase III NANP and blood-stage antigens, including MSP1 Controversial results in the field; apparently no actual protection NYVAC-Pf7 Phase IIa NYVAC virus inserted with CSP, TRAP, LSA-1 Humoral and cellular response to different components. MSP-1, SERP, AMA-1 and Pfs25 genes Delay in parasite patency but no actual protection [31] MSP-1, -2 and RESA Phase I Recombinant MSP-1, -2 and RESA in Montanide ISA720 adjuvant TRAP, CSP, AMA-1 Preclinical (rhesus) DNA vaccine with Plasmodium knowlesi TRAP, CSP, AMA-1 TRAP and LSA-1 Preclinical (chimpanzee) and MSP-1 42 (prime-boost) DNA vaccine with P. falciparum TRAP and LSA-1 (prime-boost) No adverse reactions reported in nonexposed and exposed populations One of 12 rhesus protected, remaining monkeys with lower parasitaemia [59] [104±108] [72, 73] [122] High immunogenicity [123] SERP/HRP2/MSP-1 Preclinical (Aotus) Two hybrid recombinant proteins Protection in one study, nonprotection in the other [111, 112]

Malaria Vaccines 333 and coworkers, in the 1960s, showed that an immunization procedure could induce sterile protection against malaria. In fact, the injection of irradiated sporozoites in mice [41], nonhuman primates [42] or humans [43] was shown to fully protect the host against a subsequent challenge with infectious sporozoites. Great enthusiasm was raised with these experiments at that time, boosting the search for the malaria vaccine, which was thought to be available within the following few years. Whereas irradiated sporozoites seem to be very effective inducers of protection, their use in mass vaccination is unfeasible. The main constraint is the impossibility of having access to large amounts of parasites. Research has then been directed to the development of subunit vaccines through the identification of major components of the sporozoite potentially targeted by protective immune responses. Intense research has led to the identification of the circumsporozoite protein (CSP), the major constituent of the outer membrane of the sporozoite (reviewed in [8]). CSPs of several plasmodial species share a similar structure but little sequence homology. The P. falciparum CSP has a central repeat region, presenting several repeats of the asparaginealanine-asparagine-proline (NANP) peptide, which is the major B-cell epitope within the protein. CSP is one of the main and most studied malaria vaccine candidates. Early studies were aimed to induce antibodies able to block the binding and the entrance of the sporozoite into the hepatocyte. It was shown that monoclonal antibodies against the repeats of CSP neutralized the infectivity of P. falciparum and P. vivax in monkeys [44]. Trials in humans showed that a malaria vaccine targeting the sporozoite invasion in the hepatocyte should elicit very high antibody titres, which should be continuously maintained because the sporozoite remains in the bloodstream for a very short period of time, and there is no time for an infection-induced booster-dependent effect [45, 46]. This is a major drawback in the hepatocyte invasion-blocking rationale. A reevaluation of three trials of NANP-based vaccines showed that there was no evidence for protection by these vaccines against P. falciparum malaria [47]. Liver stage The liver-stage antigens represent the second target for a malaria vaccine. During this stage, immunity is mostly mediated by cellular-dependent mechanisms involving CD8 T cells, CD4 T cells, natural killer (NK) cells and gd T cells [33]. The most important mechanism, according to the data from studies in mice, would be the production of IFN-g by activated CD8 T cells, which induces the infected hepatocyte to synthesize NO ± a potent antiparasitic activity. Other mechanisms include cytotoxicity by activated CD8 CTL through the release of perforin and granzyme B and the induction of apoptosis of the infected hepatocyte through the cross-linking of Fas ligand on the surface of the activated CD8 CTL with Fas in the infected cell membrane. Finally, antibodies can also recognize parasite antigens on the surface of the infected hepatocyte and mediate antibody-dependent cellular cytotoxicity (ADCC), especially by NK cells. The highly protective immunity induced by immunization with irradiated sporozoites is acquired only if radiation is given at sublethal doses. The attenuated parasite retains its ability to invade and partially develop inside the hepatocyte ± but not fully mature in it ± being able to elicit cellular defence. The killed sporozoite does not have this effect and induces no protection. This fact points against purified sporozoite antigens as proper malaria vaccine candidates and suggests that only those expressed during the liver stage could present efficacy. Several parasite antigens expressed during the liver stage have been identified and postulated as potential targets for a pre-erythrocytic vaccine. One of the leading antigens, again, is CSP, expressed both in the sporozoite and during the liver stage. So, much of the research involving CSP has switched from the immunodominant repeats inducing humoral response to regions able to induce T-cell responses. The malaria vaccine RTS,S/SBAS2, which conferred sterile protection in six of seven immunized human volunteers, is an engineered construction in which a portion containing the repeats and the C-terminus region of CSP was fused to the hepatitis B virus surface antigen (HBsAg), and this recombinant antigen is formulated with the AS02 (formerly SBAS2) adjuvant [48]. Although effective, no correlates of protection were identified. Despite the very encouraging results with the RTS,S/SBAS2 vaccine, it was shown that the strong immunity elicited was short lived [49, 50]. CSP is also a constituent of most multiantigen DNA vaccines in testing or to be tested in the near future. Other identified liver-stage antigens include liver-stage antigen-1 (LSA-1), LSA-2, LSA-3, SALSA and STARP, among others [33]. For some of them, such as LSA-1 and LSA-3, considerable data have also accumulated. LSA-1 is expressed specifically in liver-stage parasites, and no homologues for this antigen have been found in mouse or nonhuman primate malaria. Studies in humans associate LSA-1-specific proliferative, cytokine and antibody responses with protection [13]. The results obtained using conventional approaches were strengthened by reverse immunogenetic studies. It was shown that there was an association between expression of HLA-B53 and protection against severe malaria in Gambian children, and, by isolating and characterizing peptides bound to HLA-B53 in Gambian adults, a peptide eliciting a B53-restricted CTL response was shown to share amino acid sequence with a portion of LSA-1 [51]. These and several other data support the evaluation of LSA-1 in clinical trials, and vaccines containing LSA-1 are currently under investigation [13]. LSA-3 has been subjected to preclinical trials in chimpanzees, and results are very encouraging.

334 L. J. M. Carvalho et al. Indeed, six of nine animals immunized with different LSA-3-derived constructions were fully protected against a challenge with P. falciparum, and other two monkeys showed a delay in the patency of infection [52]. Merozoite Besides the sporozoite, the merozoite is the only stage in the human host in which the malaria parasite is extracellular. This makes it a `visible' target for antibodies, and much work in malaria vaccine development has been directed to produce antibodies effective in killing or inhibiting merozoite attachment, entrance or development in the RBC. However, it stays only for a short period time as a free organism before invading a new RBC, and this could raise some concern against the rationale for an antimerozoite vaccine. However, in contrast to the sporozoite, several cycles of merozoite release will occur during a malaria infection, making them often available. The merozoites released in the first cycle would serve as an antigen source to stimulate a memory response in vaccinated individuals, leading to a high antibody production that might counteract further development of the blood infection at subsequent cycles. Protective immunity induced by immunization with merozoites was shown more than 25 years ago with the primate parasite Plasmodium knowlesi, using Freund's complete adjuvant [53]. Several mechanisms could contribute to the parasite inhibition induced by an antimerozoite vaccine. The first obvious one would be the blocking of the merozoite attachment to the RBC. In the case of P. vivax, it has been well established that the Duffy antigen receptor for chemokines (DARC) is the receptor in the RBC for the merozoite, which expresses the Duffy-binding protein (DBP) [54]. Homozygote individuals lacking DARC are refractory to P. vivax infection. Some effort has then been undertaken to develop a P. vivax vaccine based on a construction mimicking DBP. In the case of P. falciparum, the RBC receptor for merozoites of most P. falciparum strains has been defined as glycophorin A, with specificity for sialic acid residues. A major ligand in P. falciparum is the erythrocyte-binding antigen-175 (EBA-175), located in the microneme [55]. Blocking an EBA-175-binding site inhibits parasite multiplication in vitro [56]. However, this binding site seems to be poorly immunogenic under natural transmission. In addition, some strains can invade sialic acid-deficient RBC [57], and results suggest a redundancy in the molecular interactions necessary for invasion [58]. This may have implications for a vaccine strategy designed to interfere with specific receptor±ligand interactions. Immunization of Aotus monkeys with recombinant protein, DNA or DNA, followed by immunization with recombinant EBA-175 region II, induced a significant antiparasitic effect, and some monkeys did not require treatment [59]. Several merozoite surface proteins (MSPs) have been identified, but for most of them their function is still unknown. In the case of the major MSP, named MSP-1, a role has been postulated in merozoite binding to the RBC and in the subsequent biochemical mechanisms involved in invasion [60]. This protein is synthesized as a precursor of 185±210 kda in the late schizont stage and is processed to generate several polypeptides of varied molecular weights. A 42 kda polypeptide (MSP-1 42 ) is kept attached to the merozoite membrane, and it is further processed to generate a 19 kda polypeptide (MSP-1 19 ), which is the only one that goes into the host cell [61]. MSP-1 has been the most studied blood-stage vaccine candidate antigen. Several studies have shown that antibodies against MSP-1 were able to inhibit plasmodial growth in vivo and in vitro, and MSP-1 42 and MSP-1 19 have been defined as the main targets. The latter is well conserved among parasite isolates from different geographical regions. Mice immunized with MSP-1 19 from Plasmodium yoelii were protected against infection with this parasite [62, 63]. Similarly, protection was achieved in Aotus monkeys immunized with recombinant P. falciparum MSP-1 19 [64]. However, in this case, only the yeast MSP-1 19 in association with Freund's adjuvant was able to confer protection. Similar results were obtained with Aotus nancymai monkeys, with five of seven animals immunized with MSP-1 42 in Freund's adjuvant at priming and with ISA51 at booster injections being protected [65]. In this same work, a group of animals received full-length recombinant MSP-3 in the same scheme, and six of seven animals were protected. Based on these and other results, it has been claimed that blood-stage antigens can be effective only if given with Freund's adjuvant ± a major concern, as this adjuvant cannot be used in humans. However, we have recently been able to induce protection in Saimiri monkeys using MSP-3 formulated with the AS02 adjuvant (unpublished data). AS02 has passed several clinical trials, and permission for use in humans is expected in the near future. In a Phase I trial with MSP-1 19 conjugated to tetanus toxoid, adverse local reactions were observed, resulting in recommendations for caution in the use of this formulation [66]. Besides MSP-1 and -3, at least six other MSPs have been described in P. falciparum: MSP-2 [67], MSP-4 [68], MSP-5 [69], MSP-6 [70], MSP-7 [71] and MSP-8 [69]. MSP-2 has been well characterized and undergone Phase I trials in nonexposed [72] and malaria-exposed individuals [73]. Another merozoite surface-associated antigen is the acidic±basic repeat antigen (ABRA) [74]. Proteins located in merozoite apical organelles have also been identified and proposed as vaccine candidates. The rhoptry proteins apical membrane antigen-1 (AMA-1), rhoptry-associated protein-1 (RAP-1) and RAP-2 are some of the best studied. AMA-1 has been recently postulated to play a central role in RBC invasion by the merozoite [75]. Immunization with AMA-1 or passive transfer of anti-ama-1

Malaria Vaccines 335 antibodies protected mice against lethal challenge [76]. Trials in rhesus monkeys with the P. knowlesi native AMA-1 protein conferred protection against homologous infection [77], and with a recombinant protein from Plasmodium fragile provided partial protection against homologous as well as P. falciparum infection in Saimiri monkeys [78]. RAP-1 and-2 have undergone immunogenicity and efficacy trial in Saimiri monkeys, and partial protection was observed in monkeys vaccinated with native or recombinant proteins in Freund's or Montanide ISA720 adjuvants [79]. In addition to invasion blocking or inhibition, antimerozoite antibodies may mediate other protective mechanisms such as complement-mediated lysis and act through cooperation with Fc receptor-bearing cells. Antibodies of cytophilic subclasses (immunoglobulin G1 (IgG1) and IgG3 in humans) may bind to merozoites and facilitate their uptake by phagocytes through opsonization, or mediate ADCC. Other possible mechanism is antibody-dependent cellular inhibition (ADCI) [5, 80], in which cytophilic antibodies are supposed to form a bridge between merozoite antigens and Fc receptors on the monocytes. The latter through cytokine release are responsible for a parasitostatic effect. Antibodies to at least two blood-stage antigens have been shown to inhibit parasite growth in ADCI assays ± MSP-3 [81] and glutamate-rich protein (GLURP) [82]. Infected RBC Once it has gone inside the RBC, the parasite is supposed to have found a safer place to stay. RBCs do not present MHC class I molecules on their surface, and, so, contrary to the infected hepatocyte, they cannot be attacked by CD8 T cells. In addition, parasites were supposed to be somehow hidden from antibodies, but this is not true, because: (a) parasitic antigens are expressed on the RBC surface, becoming accessible to antibodies; (b) it is believed that antibodies can penetrate the infected RBC through the parasitophorous duct and bind to the intracellular parasite [83]. In any case, mechanisms of immunity may act during this stage. CD4 T cells act through the release of cytokines that may exert parasiticidal or parasitostatic effect, activate macrophages and provide help mainly for antibody production by B cells. Antibodies against parasite molecules expressed on the RBC surface membrane could act by either allowing ADCC or complement-mediated lysis of the infected cell or by mediating opsonization by cooperation with monocytes. In the case of P. falciparum, some of these proteins have the important function of mediating the binding of the infected RBC to the endothelium of capillary vessels. The sequestered RBC provides the proper environment for the relatively safe maturation of the parasite, because, if kept circulating, the parasitized RBC could be taken by macrophages in the spleen. Antibodies blocking the attachment of the parasitized RBC to the endothelium could allow phagocytes to act and could still impede the development of cerebral malaria, as there is an association between parasitized RBC sequestration in the cerebral capillary vessels and this malaria complication. Several molecules have been identified that could elicit antibodies able to mediate the above-described phenomena. One of the most studied molecules is the ring erythrocyte surface antigen (RESA) [84]. Anti-RESA antibodies raised in mice by immunization using immunostimulating complexes (ISCOMs) as adjuvant were able to inhibit parasite growth in vitro [85]. RESA has been used in potential multiantigen blood-stage vaccine together with MSP-1 and -2 in Phase I trials [73]. The serine-rich protein (SERP or SERA) is a soluble protein expressed in the schizont stage and secreted in the parasitophorous vacuole. Epidemiological studies show that there is a correlation between the response to SERP and protection [86]. Anti-SERP antibodies seem to inhibit merozoite dispersal from mature schizonts [87]. GLURP is expressed during all stages of parasite development in the human host, including the liver stage [88]. It may therefore elicit several types of immune response, and a response against GLURP has the potential of being effective through several mechanisms acting on each stage. Epidemiological studies also show a correlation between anti-glurp antibody response and protection [89]. In addition, anti-glurp antibodies inhibit parasite growth in ADCI assays [82]. Other proteins of interest that are located on the RBC membrane are the erythrocyte membrane protein-1 (EMP-1) [90], EMP-2 [91] and EMP-3 [92]. PfEMP-1, which binds to the receptors such as CD36 in the endothelium, is a family of proteins encoded by the so-called var genes [19]. A single parasite has in its genome several (50±150) var genes coding for variant antigens. A vaccine targeting variant antigens must focus on their conserved regions. PfEMP-3 is expressed not only on the RBC surface, but also by the liver stage and by the sporozoite [93], suggesting that it may have an important role during all developmental stages of the parasite. These characteristics make it also a potential candidate for a malaria vaccine. Parasite toxins A malaria infection causes most of the symptoms in nonimmune individuals at the time of merozoite release from RBC infected with mature schizonts. These symptoms seemingly result from a wave of toxaemia owing to molecules released at that moment by synchronic parasites. Much work has been directed to identify the parasite toxins, and phospholipids have been incriminated as major candidates owing to their potent stimulation of tumour necrosis factor-a (TNF-a) production by macrophages [94]. This cytokine, in turn, has important roles not only in the protection but also in the pathogenesis of malarial disease. There is considerable evidence that TNF-a may be responsible for the malarial

336 L. J. M. Carvalho et al. fever [95], and TNF-a has been implicated in the pathogenesis of complications such as cerebral malaria [96] and malarial anaemia [97]. These data led some research groups to develop a rationale for the so-called antidisease vaccines [98]. Using this approach, most of malaria-related symptoms, as well as possibly some of the malaria complications, could be avoided. Such a vaccine is plausible but not intended to be used alone, rather in association with an antiparasite vaccine. In fact, blocking the clinical disease without clearing the parasite in a nonimmune individual might be dangerous, as a chronic infection could even lead to a life-threatening condition. Sexual stages The last possible target for a malaria vaccine is the sexually differentiated parasite stages, in an approach called transmission-blocking vaccines [99]. Antibodies developed against some proteins of the sexual stages may block the development of the parasite in mosquitoes. For instance, antibodies against the gametocytes may block fertilization [100], and antibodies against the ookinete might block the passage across the midgut [101]. Because the asexual forms would not be targeted, an individual having such antibodies would not be protected against malaria infection but would impede the parasite to be transmitted to a mosquito and, consequently, to other persons. This kind of vaccine is called `altruistic' because the vaccinated individual is not benefited; he is not protected but may protect the surrounding community by decreasing the possibility of transmission. As in the case of an antitoxin vaccine, the transmission-blocking vaccine is intended to be administered in association with an antiparasite vaccine. Several candidate antigens have been identified and evaluated. The leading candidates are Ps25, Ps28, Ps48/45 and Ps230 [99, 102]. Pfs25 has been included in multiantigen, multistage vaccines [31]. MULTISTAGE APPROACHES It has been argued that a successful malaria vaccine must combine antigens from different stages of the parasite development, inducing several mechanisms of protection. The concept of a multiantigen, multistage malaria vaccine has been put into practice for some time and has seen a major advance with the advent of DNA vaccines (see below). One of the most studied candidates is SPf66 ± a synthetic polymer that combines portions of three blood-stage antigens linked to each other by the NANP peptide derived from CSP. Despite the first trials in Colombia resulting in high protection [103, 104], other extensive field trials thereafter in Latin America [105], Africa [106, 107] and Asia [108] failed in confirming those results. A recent reevaluation of the trials conducted with SPf66 [47] showed that there was no evidence for protection by SPf66 vaccines against P. falciparum in Africa, and there was a modest reduction in the attacks of P. falciparum malaria in other regions, justifying further research with SPf66 vaccines in South America. The authors also conclude that trials to date have not been of sufficient size to evaluate the effect of malaria vaccines on mortality or on severe malaria requiring admission to hospital. A multiple antigen peptide (MAP) construction containing several T- and B-cell epitopes of CSP has been tested in mice and three monkey species [109]. The MAP construction in association with alum adjuvant presented low immunogenicity. However, the antibody titres increased in one monkey previously immunized with sporozoites, showing that the vaccine may be useful in individuals with a previous history of malaria infection in endemic areas. Moreover, the concomitant use of another immunostimulant, QS-21, resulted in improved immunogenicity. A different approach is the use of viral vectors containing genes coding for plasmodial antigens. NYVAC-Pf7 is a pox-vectored multiantigen, multistage vaccine candidate. It utilizes the attenuated NYVAC strain of vaccinia virus expressing genes coding for proteins expressed during the sporozoite (CSP and TRAP), liver (LSA-1), blood (MSP-1, SERP, AMA-1) and sexual (Pfs25) stages of the parasite's life cycle. In a Phase I/IIa trial, only one of the 35 vaccinated volunteers was protected, and, overall, there was a delay in the patency [31]. Two P. falciparum multiantigen candidate vaccines are under investigation in CDC, Atlanta. One, a 41 kda protein called CDC/ NIIMALVAC-1, contains 21 B- and T-cell epitopes from a variety of pre-erythrocytic, erythrocytic and sexual stages: CSP, LSA-1, MSP-1, TRAP, MSP-2, AMA-1, RAP-1, EBA- 175 and Pfg27. CDC/NIIMALVAC-1 has been expressed in insect cells using the Baculovirus system. The recombinant protein reacted with mouse antibodies specific for individual B-cell epitopes in the vaccine construct and with sera from clinically immune Kenyan adults. An immunization study in mice demonstrated that the recombinant protein is immunogenic, inducing high-titre antibodies as well as lymphocyte proliferation and IFN-g production [110]. These results demonstrate that individual B- and T-cell epitopes can be assembled to create synthetic genes that encode proteins capable of eliciting specific antibody and T-cell responses. A second candidate, FALVAC-2, containing MSP-1 19, Pfs25, region II of EBA-175, as well as 30 B-cell epitopes and 25 T-cell epitopes from a total of 13 stage-specific antigens, is under development. Similar approaches are underway for the development of multivalent, multistage P. vivax vaccines. The Australian Malaria Vaccine Program is developing prototype asexual-stage vaccines based on MSP-1, MSP-2, RESA, AMA-1, RAP-1 and RAP-2. Human vaccine trials have been conducted with combinations of recombinant MSP-1, MSP-2 and RESA in the Montanide ISA720 adjuvant, with the most recent being a Phase IIb trial in children in a highly endemic area of Papua New Guinea [73]. A hybrid molecule containing the blood-stage antigens